Plasma processing systems have long been employed to process substrates (such as semiconductor wafers) to produce integrated circuits. Plasmas may be generated via many different methods, such as electron-cyclotron-resonance plasmas (ECR), inductively-coupled plasmas (ICP), or capacitive coupled plasmas (CCP). In many cases, confining the plasma to a specific region within the processing chamber, such as within the region directly above the substrate being processed, may provide certain advantages.
To facilitate discussion, FIG. 1 shows an example of a low-pressure plasma reactor 100 during which plasma is confined during processing. Consider the situation wherein, for example, a substrate 124 is placed on an electrode 110, which is mounted to a pedestal 120 that is connected to a chamber 102. Electrode 110 is connected to a remote power supply 114, such as a radio frequency (RF) power generator, through the interior of pedestal 120. A processing gas 150, which may be a mixture of chemicals, may be introduced into chamber 102 through an inlet 104 when the pressure in chamber 102, which may be lowered by a pump (not shown), has reached a desirable level. To process substrate 124, electrode 110 may capacitively couple the power from power supply 114 with processing gas 150 to form a plasma 106. Usually, plasma 106 is contained within a desired region of chamber 102 by a set of confinement rings 108. During substrate processing, gases from plasma 106, which may include a mixture of chemical components from processing gas 150, chemical components formed by reactions within plasma 106, and chemical byproducts from the processing of substrate 124, may flow through confinement rings 108 and a non-plasma chamber volume 128 before being removed from chamber 102 through an outlet 126. This route is illustrated by a path 136 and usually causes the interior of chamber 102 to be exposed to highly reactive gases even when plasma 106 is contained.
However, during processing of substrate 124, plasma 106 may unexpectedly or uncontrollably migrate out of the desired region within chamber 102. In other words, an unconfined plasma 138 may form in a region of chamber 102 that is outside of confinement rings 108. The formation of unconfined plasma 138 is undesirable because unconfined plasma 138 may alter the quality of processing plasma 106 in a way that may cause at least one of the following to occur: significantly degrade the performance on substrate 124, damage chamber 102, and damage pedestal 120. For example, substrate 124 may become damaged due to a change in an etch or deposition rate and/or may be damaged by being contaminated with particulate defects or elemental contamination generated by unconfined plasma 138. Processing chamber 102 and/or pedestal 120 may be physically damaged by, for example, erosion or corrosion of chamber materials as a result of exposure to unconfined plasma 138. In addition, components of processing chamber 102 may experience electrical damage because unconfined plasma 138 may change the path by which plasma power is returned to ground through the chamber. In an example, the plasma power from power supply 114 may return to ground through chamber components which may not be designed to carry plasma power.
As can be appreciated from the foregoing, plasma unconfinement events may be caused by many different factors. For example, a plasma may become unconfined if the plasma becomes unstable. In another example, a plasma unconfinement event may occur if electrical arcing occurs within the processing chamber. In yet another example, a plasma may become unconfined if processing parameters, such as plasma power, plasma composition, gas supply flows, operating pressure, and the like, fluctuate.
Also, the occurrence of the plasma unconfinement events may be sporadic and may tend to be unpredictable. One reason for the unpredictability is that the unconfined plasma may have different forms. In addition, the specific effects that a plasma unconfinement event may have on substrate processing generally cannot be anticipated due to the variable and unpredictable form exhibited by unconfined plasma. For example, an unconfined plasma may have low density or high density. In another example, the space occupied by the unconfined plasma may be large or small. In yet another example, an unconfined plasma may be a stable plasma or may be a fluctuating, sporadic plasma. Even the location of the unconfined plasma within the reactor may change during processing.
Various methods have been employed to detect plasma unconfinement events. One method includes utilizing an electrostatic probe that usually has multiple electrodes, such as a VI probe or a Langmuir probe to detect a plasma unconfinement event. In an example, a Langmuir-style probes, which may be unprotected electrodes (usually made from metal), may be exposed to the chamber environment. Since the Langmuir-style probe are typically electrically biased such that when the probe is exposed to plasma a direct current (DC) can flow from the plasma to the electrode. For example, a Langmuir-style probe 122 is positioned within the plasma environment that is outside of the desired plasma confinement region. By employing a current detector 148, DC current changes on Langmuir-style probe 122 via a power supply 118 may be detected. Also, a DC power supply (not shown) may be employed to bias the probe.
However, the operational requirements of Langmuir-style probes (i.e. that the electrodes are unprotected and that a DC electrical contact with the plasma exist) limit the utility of the Langmuir-style probes in detecting plasma unconfinement events. Also, due to the unpredictable nature of the plasma unconfinement events, the Langmuir-style probes may have to be operating continuously while the substrate is being processed in order to be effective. However, continuous usage may result in exposing the unprotected electrodes of the Langmuir-style probes to the mixture of chemical species that is usually present in the reactor chamber during plasma processing. The mixture of chemical species, which includes chemicals supplied for processing of the substrate, new chemical species generated within the processing plasma, and chemical byproducts formed during the processing of the substrate, typically includes both corrosive components and depositing components that may detrimentally affect the ability of Langmuir-style probes to function properly. In an example, corrosive components (e.g. chlorine, fluorine, and bromine, etc.) may cause the electrodes to corrode, thereby causing the Langmuir-style probe to not function properly, such as failing to timely and/or accurately detect a plasma unconfinement event. In addition, corroded electrodes may become a source of particulate defects and/or metallic contamination that may indirectly damage the substrate being processed. In another example, the depositing components of the mixture (e.g. inorganic SiOx-based byproducts and organic CFx-based polymerizers) may result in the formation of an electrically-insulating film on the electrodes of the probe: thus, the film may interfere with the required plasma-electrode DC contact, thereby preventing the probe from accurately and/or timely sensing the presence of a plasma. As can be appreciated from the foregoing, the Langmuir-style probes may not be ideal in detecting plasma unconfinement events.
Another method that has been employed is to identify the changes in the bias voltage of a substrate during processing to detect plasma unconfinement events. With reference to FIG. 1, a bias voltage on substrate 124 may be produced when power provided by power supply 114 interact with plasma 106 within reactor 100. Typically a sensor 140 may be installed (e.g. in electrode 110) to allow direct measurement of the bias voltage on substrate 124 during processing and a bias voltage detector 144 may be employed to compare the bias voltage against a threshold. Thus, when the characteristic of plasma 106 is altered due to unconfined plasma 138, sensor 140 may be employed to measure the bias voltage and bias voltage detector 144 may be employed to detect the changes in the bias voltage.
Additionally or alternatively, a change in the bias voltage may be indirectly detected by measuring changes in parameters which are related to the substrate bias voltage. For example, when the substrate bias voltage changes because of unconfined plasma 138, the power supplied by power supply 114 to electrode 110 to maintain plasma 106 may also change. Therefore, monitoring the power supplied to plasma 106 with a RF power detector 142 may allow detection of plasma unconfinement events.
However, the utility of monitoring bias voltage to detect plasma unconfinement events is limited by the difficulty in detecting changes in the bias voltage caused by plasma unconfinement events. Detecting changes in bias voltage is particularly difficult when higher frequency generators (such as 60 MHz) are utilized to generate a plasma. Bias voltage generated by higher frequency generators are small and since a plasma unconfinement event usually occur at lower power levels, differentiating the plasma unconfinement event from the changes in the DC bias signal may be difficult or impossible to detect. Therefore the utility of this technique is limited because of the inability to reliably detect plasma unconfinement events.
In yet another prior art approach, an optical sensor may be used to detect plasma unconfinement events. Those skilled in the art are aware that plasma generally emits light. Thus, an optical sensor may be employed to detect the light emitted from an unconfined plasma. In an example, with reference to FIG. 1, an optical sensor 132 may be installed adjacent to a transparent window 130 with a line-of-sight into an area of chamber 102 in which monitoring is desired (denoted here as a passage 134). Thus, when plasma 106 becomes unconfined plasma 138, light from unconfined plasma 138 may enter passage 134 and may pass through window 130 to be detected by optical sensor 132. Upon detecting the light, optical sensor 132 may send a signal to an optical signal detector 146. If the signal is above a pre-defined threshold, optical signal detection 146 may provide an alert indicating that unconfined plasma 138 has been detected.
However, employing the optical sensors to detect plasma unconfinement events may be limited because detecting the light emitted from unconfined plasma 138 may be difficult because the light emitted by unconfined plasma 138 is significantly dimmer than the light emitted from processing plasma 106. In addition, the positioning of optical sensor 132 outside of chamber 102 may make “seeing” the light difficult through transparent window 130 since the reactive chemicals may cause transparent window 130 to be less than transparent. In other words, the reactive chemicals may cause a layer of films to be deposited on transparent window 130, thereby significantly reducing the amount and/or quality of light that be detected by optical sensor 132. Furthermore, the utility of optical sensors is dependent upon having viewing access into the processing environment. However, placing windows and/or viewing passages in all locations that may have to be monitored may not always be feasible.
Although various methods have been implemented, each of the methods does not provide a dependable and comprehensive solution for detecting plasma unconfinement events. In an example, the Langmuir-style probe is susceptible to corrosion, which may negatively impact the probe effectiveness in identifying plasma unconfinement events. In another example, depending upon identifying changes in the bias voltage to determine plasma unconfinement event may be dependent upon differentiating the changes in DC bias signal from a plasma unconfinement event. The ability to differentiate the two may be difficult when the DC bias signal is being generated by a higher frequency generator (such as 60 MHz) while the unconfined plasma is occurring at a lower power level. In yet another example, utilizing optical sensors to detect light emitted from unconfined plasma is limited by the availability of viewable passage and/or the inability to detect the light due to the obstruction that may cause the viewable passage “unviewable”.